In July 2026, the quantitative trading industry witnessed its most devastating fortnight on record.
The Mathematics of Leverage Destruction
Leopold wasn’t a novice. He wrote what was called the “AI boom bible” and was ahead of the market on AI infrastructure. But being right about direction doesn’t mean you survive long enough to collect.
The fatal flaw: **leverage’s nonlinear destruction effect.**
“` Leverage Collapse Formula: Account blow-up condition = Asset decline ≥ 1 / Leverage ratio
4x leverage → 25% asset decline = margin call SA portfolio actual decline: 35%-47%
Remaining capital = Initial × (1 – Leverage × Asset decline) = 1 × (1 – 4 × 0.40) = -0.60 (Capital wiped out + still owing) “`
This is why SA went from $45B to $100B — not “lost some money,” but had their public market positions effectively zeroed out after forced liquidation by prime brokers, retaining only $5B in Anthropic private equity.
Jupiter faced the identical math: 4x leverage means a 25% decline triggers liquidation, and their holdings dropped 43% in July.
**The core lesson: Leverage isn’t a return amplifier — it’s a survival time compressor.** During periods when your thesis is correct, leverage magnifies gains. During periods when you’re wrong, leverage eliminates your chance to recover.
Why CTA Strategies Outperformed
While AI quantitative funds imploded, domestic data revealed a striking divergence:
| Strategy | July Drawdown | YTD Return | % Positive |
|---|---|---|---|
| Equity Quant | -9.52% | +0.74% | 17.35% |
| Quantitative CTA | -1.80% | +5.98% | 76.60% |
CTA strategies’ advantage isn’t about earning more — it’s about **surviving longer**:
- **Multi-asset coverage:** Precious metals, agriculture, energy, government bonds — no single-market dependency
- **Bidirectional:** Trend-following strategies profit in both up and down markets
- **Natural leverage constraints:** Commodity futures margin requirements inherently limit excessive leverage
- **Crisis Alpha:** Trend signals become clearer during volatility spikes
Simply put: SA and Jupiter bet on one direction (AI up) + one asset class (semiconductor stocks), while CTA operates on “if not here, then there.”
Three Formulas That Could Have Prevented These Blowups
Formula 1: Maximum Leverage Cap
“` Max Leverage = Maximum acceptable drawdown / Maximum expected asset volatility
Example: 20% drawdown tolerance, 50% max expected volatility Max leverage = 20% / 50% = 0.4x (no leverage)
With 4x leverage, a mere 5% decline triggers your 20% drawdown limit “`
Formula 2: Dynamic Position Sizing (Inverse Volatility Weighting)
“` Position size = Account risk budget / (ATR × Contract multiplier)
Example: $50,000 account, 1% risk per trade ($500) ATR(14) = $30, Multiplier = 100 Position = $500 / ($30 × 100) = 0.17 lots “`
Core logic: **higher volatility → smaller position.** When market volatility doubles, your position automatically halves. Regardless of how wild the market gets, your single-trade loss stays within budget.
Formula 3: Portfolio-Level Drawdown Circuit Breaker
“` Circuit Breaker Triggers: Daily drawdown ≥ 2% → Reduce positions by 50% Weekly drawdown ≥ 5% → Flatten all positions, observe Monthly drawdown ≥ 10% → Stop trading, full system review
Recovery: 5 consecutive trading days without new lows → Restore 50% positions “`
This is the most overlooked yet most critical formula. Neither SA nor Jupiter had this mechanism — their leverage didn’t auto-shrink during losses. Instead, margin calls forced them to liquidate at the worst possible moment.
Leopold’s Promise: What “Permanent Deleveraging” Really Means
On August 1, Leopold issued a formal apology letter to investors with three commitments:
- Permanently remove all leverage from the balance sheet
- Never borrow money to amplify positions again
- Operate solely with proprietary capital going forward
Translation: **From 4x leverage to zero leverage. From $45 billion to $100 billion.**
This transformation is a microcosm of the quantitative industry — when the market casts its vote, all the “smart strategies” in the world matter less than a single risk management formula.
The irony? Citadel acquired SA’s $16 billion stock portfolio at a 10%+ discount. Ken Griffin made this bet because he had the risk infrastructure to absorb these positions. As an independent trader, you need your own “risk infrastructure” — just scaled appropriately.
The Bottom Line
This wave of AI quantitative fund blowups isn’t proof that “AI doesn’t work.” It’s proof that **leverage + concentration + one-directional bets** don’t work.
The quantitative approach itself isn’t flawed. What’s flawed is arrogance toward risk. Whether you manage billions or trade your personal account, the market enforces one rule equally for everyone: **Survive first. Then talk about returns.**
*Looking to experience professional-grade quantitative copy trading? [ECMarkets](https://i.ecmarkets.com/api/client/pm/2/793EN) offers MAM/PAMM accounts with automated strategy execution, minimum deposit just $1,000, and 30% commission rebate for new accounts. Connect: Telegram @DongyiTrade*
